`Intervention: Work at Spotify!`

Is it possible to determine—or at least tentatively predict—where Spotify is heading in the near future? If Spotify’s IPO in April 2018 was both a success (at least in terms of investors’ economic faith) as well as an uncertain forecast for the future of music (can Spotify really contine to grow?), similar questions around the fate of the company have been posed for years. Such speculations have been driven by the paradox that the music industry needs streaming services like Spotify just as much as the streaming business relies on record labels. However, there have also been less speculative propositions. A year before the IPO, in April 2017 for example, the site Zatz Not Funny! imagined it was possible to determine Spotify’s trajectory and presented a number of recent job announcements from Spotify to substantiate its claim that the company was on an “unexpected foray into hardware.”¹ Several job listings—“Senior Product Manager-Hardware,” “Product Manager-Voice” and “Director of Product, Natural Language Understanding”—seemingly provided clues that Spotify was working (or would work) on developing its own music hardware. “It seems safe to assume that, in some fashion, Spotify is interested in making a device that’s capable of playing music,” the Verge confidently reported.² Based on a series of job listings, it was deduced that Spotify was following in the footsteps of many of its tech siblings, who had recently started to produce their own tangible computer products.³

Even if these speculations were based on actual job listings, they turned out not to be true. The job postings managed to trigger the imaginations of the tech press and were used as a source for generating prophecies about Spotify’s future advancements—which in the end did not happen. These days, however, job-based speculation is part of a broader trend in labor market intelligence that involves using large collections of job listings to study local and global hiring tendencies—with the goal of potentially offering better recruiting opportunities or even predictions about where markets are moving. The company Wanted Analytics, for example, boasts of being a leading provider of real-time employment market analytics and is said to maintain a database of more than one billion unique job listings from 150 countries.⁴

In our final intervention, then, we were inspired by such efforts to “read” markets and companies through job ads. We decided to scrape our own data set of job postings from Spotify to speculate about the company’s prospects for the coming years. We knew that job listings would neither give clues regarding the imminent IPO nor would they provide any substantial evidence about Spotify’s general commercial strategy, which—as the recent IPO made clear—would be focused on growth and scale rather than on becoming a profitable company. Nevertheless, Spotify job listings became an empirical source of information about the service, and we have occasionally used them as such throughout our book. Our scraped data set of job listings, in particular, gave us explicit insights into Spotify’s corporate ideology and organization. In addition, we wanted to learn more about Spotify’s business lingo and how the company “speaks.” While analyzing the job postings, we were particularly interested in the kinds of spaces that Spotify carves out for its human employees, as compared to the machines and algorithms that form the core of the company. Job postings, we argue, represent an undervalued source of insight into media industries. As empirical evidence, they provide important clues about how corporations organize, allocate resources, and both perceive and brand themselves.⁵ So in order to map Spotify’s current and future priorities, we decided to monitor the company’s recruitment history.

In detail, the interventionist exercise resulted in the collection of some 800 job postings from Spotify’s global website between November 2016 and April 2017.⁶ A number of these job listings were duplicates or slight variations of one another, and after removing such ads, we were left with 563 unique Spotify job listings from all around the world. Such listings are not only crucial for a company in the midst of a rapid global expansion—they also provide fodder for publicity stunts. For instance, in January 2017, shortly before Barack Obama’s tenure as president of the United States ended, Spotify CEO Daniel Ek tweeted, “Hey @BarackObama, I heard you were interested in a role at Spotify. Have you seen this one?” Ek added a link to an alluring posting for the position “President of Playlists.”⁷ Ek’s tweet was in response to a remark that Obama had allegedly made to Natalia Brzezinski, the wife of the former US ambassador to Sweden, at a White House reception. In an Instagram post,⁸ Brzezinski recalled Obama joking, “I’m still waiting for my job at Spotify … Cuz’ I know y’all loved my playlist.”⁹ While Ek’s job posting did not mention Obama by name, it required a curriculum vitae that narrowed the talent pool to essentially one person with “at least eight years’ experience of running a highly-regarded nation.”¹⁰ Thus, job listings are not only vehicles for attracting qualified employees but can also be tools for improving a corporate image. Ek’s tweet was liked more than fourteen thousand times and retweeted more than seven thousand times.

Upon closer inspection, we found that Spotify had divided the job postings into three broad categories: business, technology, and product. Among the announced jobs, nearly 60 percent belonged to the first category and nearly 30 percent to the second. Thus, while Spotify may still frame itself as a tech company, business was by far the largest category in our data set for which Spotify sought new employees. Overall, our collected job positions could be traced to thirty-four different cities in eighteen countries, from Bogotá, Colombia, to Taipei, Taiwan, to Antwerp, Belgium. The majority of positions that Spotify sought to fill, however, were located in only two cities: New York (about 37 percent of the postings) and Stockholm (about 25 percent). Half of all listed positions were in the United States, with an additional 27 percent located in Sweden. By contrast, only 1 to 2 percent of the positions were in countries such as Germany, Brazil, and Japan. In other words, even though Spotify appeared to be increasing or at least maintaining its workforce in several of its satellite offices, the bulk of the positions it sought to fill were located in either the United States or Sweden.

The positions in business, technology, and product were divided into further subcategories. For instance, the business category comprised positions in content, sales, finance, label relations, and content & PR. Technology, in turn, included jobs in areas such as IT, software, infrastructure, and mobile, and the product category boasted jobs in product development, analytics, and design & user experience. The wording here is interesting to consider. Based on Spotify’s own job categorization, the product that the company is essentially selling could be narrowed down to user experiences (i.e., encounters with cultural artifacts), design (i.e., an attractive and well-functioning interface), analytics and user research (i.e., big data studies concerning customer behaviors), and operations and growth (i.e., Spotify’s speculative promise to generate potential profit in order to attract venture capital, as discussed in chapter 1). Some subcategories also fit into more than one main category, such as data & machine learning (both product and technology) and analytics (both business and technology), thus indicating how some of Spotify’s positions may float between or span departments.

When trying to attract new employees for these different positions, Spotify is in a comparatively privileged position. In rankings of Sweden’s most attractive employer, for instance, Spotify usually comes second to Google and usually ranks among the top five companies.¹¹ Advanced skills are required to land almost any job at Spotify. A number of positions, for instance, require “an advanced degree, preferably a PhD” and occasionally up to “10+ years of experience.” Even more mundane positions—such as “Music Programmer” in Berlin—require substantial skills: “We are looking for a broadly experienced Editor/Music Programmer to join Spotify’s curation/programming and editorial team in Germany.… You will curate for first rate music playlist listening and programming experiences for a multitude of our moods, moments, and genres, demonstrate a passion for performance-oriented analytics.” In order to perform this work, the applicant ought to have an “ear to the ground in the music community, focusing on Germany” and at least five years of “experience in the music industry, programming music for TV, radio or other media networks, working with a broad range of music content.”

Job descriptions within the categories of business or technology are similarly demanding. “You have experience implementing machine learning systems at scale in Java, Scala, Python or similar (not just R or Matlab),” a “Machine Learning Engineer” posting states. Other job positions are more niched, such as “Head of Christian Music,” who will “make sure that all Christian Music artists see Spotify as the first choice [and] strategic partner when it comes to launching their new music.” The “Data Curator” position requires that applicants “have a deep understanding of Latin American music (strongly preferred).” And for a single week in December 2016, it was even possible to apply for a Spotify job in Los Angeles as a “Writer: What you’ll do? Develop and pitch scripts for a variety of short and medium length non-fiction video formats.”

Our job data sets allowed us to study Spotify’s corporate lingo in detail. Job postings can be approached as snapshots of corporate discourse; they tell readers how Spotify “speaks” of itself as an employer and how it pictures and perceives its (potential) employees. In this way, job descriptions reflect attempts at self-branding, as well as expectations of future employees. Most job listings, for example, featured a company assertion—underneath the actual job description, often in italics or bold—stating that Spotify is “proud to foster a workplace free from discrimination. We strongly believe that diversity of experience, perspectives, and background will lead to a better environment for our employees and a better product for our users and our creators. This is something we value deeply and we encourage everyone to come be a part of changing the way the world listens to music.” Following company jargon, diversity among employees becomes both a desired quality among personnel and a commercial concept for making better products.

Spotify’s assertion regarding diversity is but one of many semantic examples of how job listings shed light on the narratives that the company creates about itself and its labor force. While some media research has gone into studying the ways in which users and audiences are engaged in different kinds of digital (and even unpaid and exploitative) labor,¹² fewer scholars have focused on the working environments of those who build and maintain today’s digital services. There are some exceptions, such as a number of studies conducted within the field of corporate ethnography, with anthropologists as “company insiders” who examined product design and branding, the use of information technologies, workplace practices, and the culture of the organization itself.¹³ While our collected job ads say little about the actual situation of Spotify’s workers, they do tell us something about the kinds of work they are expected to perform.

In trying to hire people, Spotify resembles other tech startups with a hipster-business appeal. “We’re pioneers,” the “Backstage” page on spotifyjobs.com proclaims. “Our industry sector didn’t exist before we arrived. Nowadays, when people think music streaming, they think Spotify. And we’re not done yet, nowhere near.”¹⁴ We were thus not surprised to find out that the three most common nouns that appeared in the job postings data set were team, experience, and product. All three indicate how Spotify perceives its corporate organization (as made up of collaborative teams), what main trait it values in its employees (experience), and what the company perceives itself as doing (building and maintaining a product).

Following these three terms, word frequencies differ somewhat between countries. Comparing the two countries where most of the jobs were located—the United States and Sweden—we could for example see that skill was ranked as the sixteenth most common noun in US job postings, while in Sweden, it was only ranked as number thirty-three. Similarly, ability was ranked as the twenty-third most common noun in the US postings but only number forty-five in the Sweden. This does not necessarily imply that Spotify considers it less important that its Swedish employees are able or skilled. It does indicate, however, that considerable regional differences exist in terms of how Spotify expresses its desire for particular skills and personality traits from its employees. Spotify is not recruiting in the same way everywhere; the company implements its own global division of labor.

Analyzing word frequencies also provided us with indications of how such divisions of labor are allocated. Focusing again on the two countries where Spotify wanted to hire the most people (Sweden and the United States), we could detect some noticeable differences. In the United States, for example, marketing was the twelfth most frequently occurring noun, but it was only ranked eighty-seventh in the Swedish case. On the other hand, software was the twenty-fourth most frequently occurring noun in the Swedish postings but only came in eighty-third in the United States. In Sweden, IT was also ranked as number fifty-three, but it did not even make it onto the short list of the one hundred most commonly used nouns in the United States—much like sales, which was the sixty-third most commonly used noun in the US postings but did not appear at all on the Swedish short list.

Judging from our collected job listings, Spotify’s sales and marketing operations seemed to be primarily located in the United States, while more technical tasks such as IT and software engineering were run from the Swedish offices. This hypothesis was also backed up by an overview of exactly where Spotify’s three large job categories (business, product, and technology) were located, as shown in figure 5.3.

The table shows that a majority of Spotify’s business-oriented positions were located in the United States, while Sweden hosted a comparatively large portion of tech-oriented jobs. It also becomes apparent that all of the open positions in countries such Norway, Italy and Mexico were on the business side. Job openings in the product and technology sectors were only available in two other locations besides the United States and Sweden: the United Kingdom and Singapore.

When Spotify describes the personality traits of its employees and/or the work atmosphere, the company most frequently uses verbs such as lead, change, develop, value, believe, listen, encourage, build, and foster. Moreover, some of the most commonly used adjectives were better, strong, free, proud, new, global, creative, and excellent. A majority (about 50 percent) of the scraped job listings related to higher positions within the company, such as director, manager, or vice president. This is surprising, given that one would expect Spotify to headhunt for more senior job placements, instead of relying on applications from open job listings. Apart from that, Spotify was looking for a comparatively large number of software engineers, specialists, planners, analysts, coordinators, product owners, and data scientists. More low-level jobs, such as assistant jobs and office coordination positions, were much less numerous.

If the site Zatz Not Funny! could predict that Spotify would venture into the domain of music hardware based on three job announcements, what can our data set of 563 job listings ultimately tell us about Spotify’s future? First, we noted a range of practical and material issues concerning Spotify’s rapid growth. For example, several job positions—particularly those in New York City—addressed the housing market. “Spotify will expand its U.S. headquarters and relocate to 4 World Trade Center and create 1,000 new jobs, according to New York Governor Andrew Cuomo,” Business Insider reported in February 2017.¹⁵ A couple of months later, in May 2017, Spotify was hiring a “Senior Real Estate Analyst” who would “lead the forecasting and planning process” of the company’s “global real estate portfolio,” spending 30 percent of his or her time on global traveling in order to “monitor and track large office space built out projects.” A similar ad for a job in “Real Estate Strategic Planning” included tasks such as creating “short and long term occupancy scenarios” and assisting “in the management of lease transactions and executed lease documents.” For this position, Spotify requested a “solid business judgement.” The fact that Spotify was in dire need of expertise in the New York real estate market suggested that the company planned to move headquarters to the United States, which did not turn out to be true.

If the New York housing speculations based on our scraped job listings turned out negative, another aspect of Spotify’s growth that we found striking—and a surmise more accurate—was the relation between humans and machines in the job listings, hinting at the use of artificial intelligence and other computational methods to give music customers the best music-discovery experience. In thirty-four of the job postings, the notion of “machine learning” was present in different ways. Hence, although Business evolved into the major organizational category at Spotify, its tech identity was still its corporate trademark.

Spotify has frequently branded itself as a software-driven company that relies on machine learning and algorithmic data mining in order to generate music recommendations and create music experiences. While streaming services such as Apple Music have taken public pride in their humanly curated playlists, Spotify has instead hallmarked itself as the techy alternative—vividly illustrated by the catchphrase, “music + math = epic,” presented in a slideshow by Ching-Wei Chen and Vidhya Murali (both employed in Spotify’s machine learning department).¹⁶

Spotify’s aim to hire people to teach machines is a frequent feature among the job listings, hinting at a profound belief in computational capacity: “A Lead Ad Tech Machine Learning Engineer to help us build a modern and dynamic ad stack which can efficiently serve ads to our [listeners]”; “a Machine Learning Engineering Manager (Chapter Lead) to drive the machine learning and data engineering practice within the Revenue mission”; “a data engineer to help us build data-driven solutions”; “We are hiring software engineers who are very enthusiastic about data to focus on building structured, high-quality data solutions”; and so on. Spotify’s apparent desires to automate and let machines handle its business is also shown in many of its startup acquisitions. In the spring of 2017, the company Sonalytic was acquired for to its sophisticated audio feature detection, as was the much-hyped French company Niland, which specializes in artificial intelligence.¹⁷

Yet despite its focus on machine learning and software-aided music curation, Spotify is of course also in need of human bodies and human labor. Even if machine-learning algorithms are capable of performing a large amount of work on their own, they are not capable of designing, monitoring, or directing themselves. Then again, the many job positions regarding machine learning and company acquisitions of cutting-edge tech companies are a strong indication that, in the near future, both machines and computational solutions will have an increased presence at Spotify. When looking at job listings within the general category of Technology, positions in the Software Engineering and Data & Machine Learning subcategories were common in our data set.

Finally, we were not able to perceive any significant changes or alterations—that is, a potential increase (or decrease) in certain Spotify job subcategories—over the half year we scraped our listings. On the contrary, when speculating about Spotify’s future, it should be stressed that a number of positions were always listed during the six-month period. It seems as though Spotify is always looking for, say, skilled software engineers—and probably has been for the past decade. Even within the general category of business, a nearly equal number of jobs were being offered within the subcategories of subscription business, finance, and sales. One conclusion from this final intervention is that job listings can give some indications and evidence of where a company is heading. A larger, longitudinal data set is needed, however, especially since a more precise forecast would have to take into account job alterations over time—that is, how the positions themselves fluctuate.

Intervention: Work at Spotify!

Notes

`Intervention: Work at Spotify!`

`Notes`